An owner asked me last year whether he should put call recording on his sales team. He had two salespeople, a reasonable close rate on one and a poor one on the other, and no idea why.
That is a good reason to look at this. What he described next was the problem. His plan was to install it quietly and see what he found out.
What these tools actually do
Strip the product marketing away and the job is simple. A tool joins your video calls, or connects to your phone system, and records them. It turns the audio into written text. Then it produces a short summary of what was discussed and what was agreed, and it can push that summary into your CRM so the record fills itself.
Most of them add some analysis on top. How long each person spoke. Which topics came up. Whether price was discussed and when. Some try to rate the call.
The transcription is good now. On a clear call with one person per microphone it produces text you can read without irritation. It stumbles on heavy accents, industry vocabulary it has never seen, and three people talking over each other in a meeting room on one laptop.
The summary is the part that saves real time. A salesperson who spends ten minutes after every call typing what happened gets that time back, and the record is more complete than what they would have written. That alone is worth the subscription at most companies, before any coaching enters the picture.
Telling people you are recording
Ask a lawyer about your specific situation rather than relying on a blog post, including mine. What I can tell you is the practice to follow, which is simpler than the law and safer than it.
Tell people you are recording. Say it at the start of the call, in plain words, and give them a way to decline. Something like: I have a notetaker on this call that writes a summary for me, let me know if you would rather I turn it off.
Most people say it is fine. A few ask you to turn it off, and you turn it off. That is the whole interaction, and it takes eight seconds.
Some tools announce themselves automatically. Some show a visible label in the meeting. Both help, and neither replaces you saying it out loud, because a small on-screen notice is easy to miss and it does not feel like being asked.
Two situations deserve more care. Calls with people in other countries, where the rules differ from the ones you are used to. And any business with professional obligations of its own, where client confidentiality may restrict what you are permitted to record at all. A law firm, an accounting practice, and a healthcare business each have a different answer here. Check yours before switching anything on.
One more thing that gets missed. If recordings and transcripts contain customer information, they are customer records, and they sit wherever the tool stores them. Find out where that is, how long they are kept, and who inside your company can open them. Those questions belong in your rules for using AI tools rather than in somebody's head.
The surveillance problem is the real risk
Back to the owner who wanted to install it quietly.
I told him that the day a salesperson discovers their calls have been recorded without being told, he loses something he will not get back. Whatever he learns from the transcripts costs him more than it returns.
Here is the thing about a recording tool. The technology is identical whether you are coaching somebody or checking up on them. The difference lives entirely in how it is introduced and what happens with what you find.
Reps work out which one it is within a month, and they are rarely wrong. The signals are obvious. If the first time anyone mentions a recording is in a conversation about underperformance, it is a monitoring tool. If a manager quotes a line back from three weeks ago to win an argument, it is a monitoring tool. If nobody ever discusses a call that went well, it is a monitoring tool.
Once a rep decides it is monitoring, the calls change. They become careful. They stop making the judgement calls that close deals, because a careful call is a safer call to have on record. You end up with a team that sounds compliant and sells less.
So introduce it properly. Say what it is for. Say who can see the recordings, and make that a short list. Say what you will do with what you find, and then do exactly that. Give the reps access to their own calls, which is the single strongest signal that this is a tool for them.
Then prove it in the first month by coaching something that went right.
What to look for in a transcript
The mistake I see next is an owner who reads everything, learns nothing, and quietly stops opening the tool by week six. Fifty calls a week is too much material. You need a reason to open a specific call.
Four things are worth reading for.
The talk ratio on discovery calls. Most tools show this automatically. A first call where the salesperson speaks for most of it is a call where nobody learned anything about the buyer. This single number finds more coachable problems than anything else in the tool, and you can act on it without reading a word of the transcript.
The moment price came up, and who raised it. Search the transcript for the price discussion and read the two minutes either side. You learn whether your rep introduced the number defensively before any value was established, and you learn what they said when the buyer pushed back. That is the most expensive thirty seconds in most sales calls.
Questions the buyer asked that went unanswered. These are easy to spot in text and almost invisible live. A buyer asks about implementation time, the rep answers something adjacent, the conversation moves on, and the question quietly becomes the reason the deal stalls.
Calls on deals that were lost. Go back to the first call after a loss. Reading the discovery conversation with the outcome already known is the most useful hour a sales manager can spend, and it usually shows the deal was never qualified. This pairs naturally with a win loss review.
Notice that three of the four are about discovery rather than closing. That is where small company sales problems usually live, and better questions on the first call fix more than any objection handling script.
The coaching conversation that follows
A recording nobody discusses is a filing cabinet. The value arrives in the twenty minutes afterwards.
Run it as a regular slot rather than an event. Thirty minutes, once a week, one call. The rep picks it as often as you do, which changes the feeling of the meeting completely.
Play the section rather than describing it. People argue with a description of what they said and accept a recording of it. Then ask what they were thinking at that moment, before offering any opinion, because the answer is often reasonable and changes your advice.
Pick one thing to change. A rep given four improvements changes none of them. A rep given one, with a specific replacement sentence to try, will use it on the next call.
And coach the good calls too, which is where most managers fail. When a call goes well, most teams say nothing and move on, so nobody learns what worked. Playing two minutes of a strong call to the whole team teaches more than any training session, and it costs nothing.
What these tools are bad at
They do not understand your business. A summary will record that pricing was discussed and miss that the buyer mentioned a budget cycle that makes a signature impossible until April. The consequential detail is often the one the model has no reason to weight.
The automated scores are worth little. A number out of ten on a sales call, produced by a model that does not know your product, your market, or what a good outcome looks like for that account, is a number you should ignore. Use the tool for the transcript, the summary, and the talk ratio. Treat the rest as decoration.
They also miss what was not said. A call can be perfectly executed and the rep still failed to ask the one question that mattered. A transcript shows you what happened, and reading for absence is a skill you bring, not one the tool provides.
And they cannot tell you why somebody bought. The reason often lives in a conversation you were never part of, inside the buyer's company, between two people you have never met.
Does a three person team need this
Often no, and I say that as somebody who likes these tools.
If you are the owner and you run most of the sales calls yourself, the coaching case disappears, because you cannot coach yourself from a transcript. The note taking case still stands. If ten minutes of typing after every call is the thing you keep skipping, and your CRM is empty as a result, a summarising tool solves a real problem and the notes-into-CRM workflow is worth setting up on its own.
With one or two salespeople, sitting in on a call occasionally costs nothing and teaches you more, because you can ask about the pause afterwards. Buy the tool when you have three or more sellers, when you cannot attend enough calls to know how they sound, or when you are about to hire and want a library of real calls for training.
Onboarding is the strongest case of all. A new salesperson who can listen to ten real calls, including two that were lost, reaches useful speed faster than any amount of product training.
The takeaway
Tell people you are recording, every time, out loud. Give reps access to their own calls. Decide before you start whether this is a coaching tool, then behave that way for long enough that the team believes you.
Read for four things rather than everything: talk ratio on discovery, the price moment, unanswered questions, and the first call on deals you lost. Then hold one thirty minute conversation a week and change one behaviour at a time.
If you sell most of the work yourself and have two salespeople, buy it for the note taking and skip the coaching features until the team grows.